Short-term Cognitive Networks, Flexible Reasoning and Nonsynaptic Learning

Abstract

While the machine learning literature dedicated to fully automated reasoning algorithms is abundant,the number of methods enabling the inference process on the basis of previously defined knowledgestructuresisscanter.FuzzyCognitiveMaps(FCMs)arerecurrentneuralnetworksthatcanbeexploitedtowards this goal because of their flexibility to handle external knowledge. However, FCMs suffer froma number of issues that range from the limited prediction horizon to the absence of theoreticallysound learning algorithms able to produce accurate predictions. In this paper we propose a neuralsystem namedShort-term Cognitive Networksthat tackle some of these limitations. In our model, usedfor regression and pattern completion, weights are not constricted and may have a causal nature ornot. As a second contribution, we present a nonsynaptic learning algorithm to improve the networkperformance without modifying the previously defined weight matrix. Besides, we derive a stopconditiontopreventthealgorithmfromiteratingwithoutsignificantlydecreasingtheglobalsimulationerro